AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

Spatial Representation Distillation and Knowledge Routing for Vision-Language-Action Models

Jungin Park, Chaoran Zhu, Changjae Oh, Kwanghoon Sohn

NeurIPS 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Spatial reasoning in VLMs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes spatial representation distillation and knowledge routing to inject 3D spatial knowledge into vision-language-action models without explicit 3D inputs.

Problem

VLAs need 3D spatial understanding for robot manipulation, but explicit 3D inputs (depth, point clouds) are costly and implicit spatial injection into visual features is limited.

Approach

Distills spatial representations into the VLA's visual features and uses a knowledge routing mechanism to direct spatial knowledge to the action pathway; based only on the abstract, the precise mechanism is not confirmed.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add attention response distillation from VGGT to the LoRA vision branch (ablation)SPARK-VLA (no SigLIP) · LIBERO averagesuccess rate (%): 58.7 -> 94.1Improved
Add knowledge router (ablation)SPARK-VLA · LIBERO averagesuccess rate (%): 94.1 -> 98.5Improved
Use SigLIP together with DINOv2 (ablation)VLA with vision LoRAs + projector · LIBERO averagesuccess rate (%): 58.7 -> 72.4Improved
Train projector from scratch in addition to LoRAs (ablation)LoRA-only VLA · LIBERO averagesuccess rate (%): <5 -> 72.4Improved
Use attention response distillation instead of feature matching (ablation)SPARK-VLA · LIBERO-Goalsuccess rate (%): 94.8 -> 98.0Improved
Use attention response distillation instead of attention-map matching (ablation)SPARK-VLA · LIBERO-Goalsuccess rate (%): 96.4 -> 98.0Improved
Add geometric distillation instead of action loss only (ablation)SPARK-VLA · LIBERO-Goalsuccess rate (%): 81.6 -> 98.0Improved
Use action-driven router instead of direct token concatenation (ablation)SPARK-VLA · LIBERO-Goalsuccess rate (%): 94.0 -> 98.0Improved
Use router instead of FiLM fusion (ablation)SPARK-VLA · LIBERO-Goalsuccess rate (%): 94.8 -> 98.0Improved
Use momentum moving average (0.9) for router targets (ablation)SPARK-VLA router · LIBERO-Goalsuccess rate (%): 94.0 -> 98.0Improved
Increase momentum beyond 0.9 to 0.9999 (ablation)SPARK-VLA router · LIBERO-Goalsuccess rate (%): 98.0 -> 97.0Hurt
Raise routing ratio rho from 0.7 to 1.0 (keep all tokens) (ablation)SPARK-VLA router · LIBERO-Goalsuccess rate (%): 98.0 -> 94.0Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 4 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy